PDSF Home

This page contains a series of graphs that use data from the PDSF batch scheduler (LSF) to give an idea of cluster utilization. All graph s show average values. The graphs were generated with RRDTool, and are updated on a roughly 15 minute interval.

This is an up to date graph of aggregate cluster CPU absolute usage according to lsload:

This is a graph of LSF Wait Queue sizes, in terms of number of jobs, according to qstat, for the last day, updated every 15 minutes (see the reasons for pending below) :

This is a graph of LSF Run Queue sizes, in terms of number of jobs, according to qstat, for the last day, updated every 15 minutes:

This is a graph of the number of running jobs by group according to bhpart, for the last day, updated every 15 minutes:

This is a graph of the number of jobs pending by reason, for the last day, updated every 15 minutes. This is generated by examining the pending reasons for each job on the various hosts. Some assumptions are made to determine the reason for the job not running.

This is a link to a list of graphs generated every 24 hours, based on individual host CPU utilization reported by the lsload command

body_close(); ?>